C20
Model Packaging and Inference
4 lessons3 hoursPlannedFree
Outcome
Export, quantize, and benchmark a model for inference.
Prerequisites
- Python basics.
Lessons
- L01Serialization and model formats · coming soonExercise: Save and reload a model.
- L02ONNX export and verification · coming soonExercise: Match outputs between PyTorch and ONNX.
- L03Quantization (INT8, FP16) · coming soonExercise: Measure accuracy loss and speedup.
- L04Benchmarking correctly: warmup, percentiles · coming soonExercise: Report p50 and p95 latency.